NCE30H12K — 30V/120A N-Channel Power MOSFET (TO-252-2L)

What Are the Technical Specifications of NCE30H12K?

The 120W dissipation rating and the 1.25°C/W thermal resistance are the same statement twice: (175 − 25) / 1.25 = 120W. That arithmetic holds together, and it also tells you the ratings table is built entirely on case temperature.

So is the 84A row. Work it backwards and the sheet's assumed hot resistance appears: (175 − 100) / 1.25 = 60W, and 60W / 84A² = 8.5mΩ. That's the number the manufacturer expects the die to have at temperature — 1.9 times the 25°C maximum.

The third row is a first-order derivation, not a printed spec: hold a 40°C/W board at 25°C ambient and the self-heating equation lands near 19A with a 125°C junction. A 65°C/W board halves that margin again.

Both currents burn within 8% of each other — 64.8W against 60W. The difference is where the heat leaves: one point assumes a case pinned at room temperature, the other assumes a case at 100°C. The line between them is just (175 − T_C) / 1.25.

When Should You Use (and NOT Use) the NCE30H12K?

✅ Use NCE30H12K when:

• Your gate driver actually runs at 10V. The single R_DS(on) row is specified at 10V and 20A. Give it that and the 4.5mΩ maximum holds.
• You have real copper under the tab. A 12–15mm thermal pad with a filled via array into an inner plane is what turns this package into a power device.
• You're replacing a TO-220 in a 12V or 24V system. 30V rating with margin for ringing, and the tab solders down instead of needing a screw and insulator.
• Burst current matters more than average current. 400A pulsed rating at a 300µs pulse width, against 84A continuous at a 100°C case.
• You're already committed to the NCE 30H family. Same footprint and pinout as the 30H10K and 30H15K, so a ladder change is a reflow change, not a re-layout.

❌ Don't use NCE30H12K when:

• Your gate drive tops out at 5V or 3.3V. There is no 4.5V row to design against. Partially enhanced is an uncharacterized state, and the temperature rise becomes yours to own. Use the 30H15K or 40H12K instead — both publish a 4.5V line.
• You need 100A continuous on a normal board. The 84A row already assumes a 100°C case, which is itself a heatsink problem. Design for tens of amps unless you have measured your own copper.
• Your load is inductive and unclamped. 350mJ of single-pulse avalanche is a real budget, but it is a 25°C-junction number. Hot, it shrinks. The 30H15K carries 1700mJ for the same footprint.
• Your rail is above 30V. Obvious, but people read the 120A and forget the 30V. The 40H12K is the same current class one voltage step up.
• You're stacking it in parallel and expecting the avalanche to share. Breakdown voltages differ between dice; the lowest-BV device avalanches first and takes the current alone.

What Are the Alternatives to NCE30H12K?

Here's the thing that costs people money: they buy the 30V part because the number is bigger, then discover their gate driver is 5V. The 30H12K is a 10V-drive device in a family where two of its three siblings publish 4.5V rows. Check the driver before the amp rating.

What does your rail actually allow?

Then the ranking flips. The 40V 40H12K has lower on-resistance than this 30V part — 4.0mΩ against 4.5mΩ at the same test conditions, with the same thermal path. Voltage class is not what sets resistance inside this family; die generation is.

Read the drawing against the sheet and one thing matters more than the pin order: pin 2 and the tab are the same net, and that net is the drain.

The tab is your switching node. Bolt it to a grounded chassis and you have shorted the supply through the device.

What Are the Typical Applications of NCE30H12K?

Battery pack disconnect and charge paths. A 30V rating fits 6-cell and 7-cell lithium packs with review, and the 4.5mΩ maximum keeps conduction loss low where the current actually flows. Keep the gate at a real 10V rail and size the copper for the average, not the peak.

Motor drive in 24V systems. E-bike controllers and small BLDC drives use this package as the phase switch. The 400A pulse rating covers commutation spikes; the 58ns reverse recovery on the body diode sets how much of that energy comes back as heat.

UPS and inverter low-voltage stages. NCE lists uninterruptible power supplies on the front page, and the 30V/120A combination is aimed at the battery-side switch. Two parts in parallel are common here — add gate resistors per device and expect the avalanche current to be unequal.

Replacing through-hole TO-220 switches. Same current class, surface-mount tab, no screw or insulator. The trade is thermal: a TO-220 bolts to a heatsink, this part sells heat only through board copper.

Why Buy NCE30H12K from ICMASS?

We sample-test NCE30H12K lots at the sheet's own condition: 4.5mΩ maximum at V_GS = 10V and I_D = 20A. That is the only on-resistance row the datasheet publishes, so it is the only one worth measuring against.

The clone market is active on this part number. VBsemi sells a NCE30H12K-VB in the same TO-252 outline, and its published rows include a low-voltage R_DS(on) line the NCE original never printed.

A part that looks better at 4.5V than at 10V is a different die wearing a similar code.

We stock the TO-252 NCE ladder together — 30H10K, 30H12K, 30H15K and 40H12K — each with its official PDF on hand, so a family swap inside the same footprint is a stock question rather than a sourcing project.

Orders ship same day from Shenzhen. Volume pricing keeps a genuine 120A TO-252 competitive against clone stock, and we will say plainly when a 40V sibling is the better buy for your rail.

Frequently Asked Questions About NCE30H12K

Q1: How much current can the NCE30H12K actually carry continuously?

A: Far less than 120A, and less than 84A on a normal board. The 84A row assumes a case held at 100°C, which on a real board means removing roughly 60W through copper alone. Hold a 40°C/W board at 25°C ambient and the self-heating equation lands near 19A at a 125°C junction. Treat 120A as a die capability, not a board capability.

Q2: Can I drive the gate from a 5V logic pin or a 4.5V rail?

A: No — this part has no 4.5V characterization at all. The only R_DS(on) row is at V_GS = 10V, I_D = 20A. The threshold is 1.0–3.0V, so a 5V drive will turn it partly on, but partly on is an unspecified resistance and the temperature rise is unmeasured. Use the 30H15K or 40H12K if your driver is 4.5V.

Q3: Why does the 40V NCE40H12K have lower on-resistance than this 30V part?

A: Because voltage class isn't what sets R_DS(on) inside this family — die generation is. The 40H12K publishes 4.0mΩ max at 10V against the 30H12K's 4.5mΩ, at the same 20A test current, with the same 1.25°C/W thermal path and the same 120A headline. It also publishes a 4.5V row that the 30H12K lacks. If your rail allows 40V, it is the better part.

Q4: Is 350mJ of avalanche energy a lot?

A: It's a real budget, but read the test condition. The sheet states L = 0.5mH, R_g = 25Ω, T_j = 25°C. Avalanche capability falls as the junction starts hot, and E_AS is a single-pulse number. For repetitive or hot avalanche, the 30H15K's 1700mJ is the family's rugged option.

Q5: What does the tab connect to, and can I screw it to a chassis?

A: The tab is the drain, and it is live. Pin 2 and the tab are the same net, so the tab swings with your switching node. Grounding it shorts the supply through the device. Solder it to a dedicated copper pad with thermal vias — never to a grounded plane.

Q6: Can I compare this part's 79nC gate charge with the 30H15K's 38nC directly?

A: Closely, yes. Both rows are measured to the same 10V gate endpoint at V_DS = 15V; only the test current differs, 20A here against 30A there. So the 30H15K really does specify about half the gate charge at the same drive voltage. What you cannot do is carry the number to a different rail — gate-drain charge scales with the drain voltage you actually switch, so re-check it at your own V_DD.

Q7: How much copper do I need under the tab?

A: A 12–15mm square pad with a filled via array into an inner plane is the practical target. Published measurements for DPAK-class parts put a bare 20mm-square pour at roughly 50°C/W board-to-air, and an optimized pad with vias at roughly 25°C/W. Beyond about 1 in² of copper the returns flatten out, so area gets you further than thickness.

Q8: Is the NCE30H12K good for paralleling?

A: It works, with two caveats. Breakdown voltages vary between dice, and in avalanche the lowest-BV device takes the current first — so the 350mJ rating is not multiplied by the number of parts you stack. Gate resistors per device are also needed to stop oscillation between gates. Share the current thermally and the arrangement behaves; assume it shares automatically and it won't.

Q9: Is a genuine NCE30H12K being sold under other brands?

A: Yes — VBsemi lists a NCE30H12K-VB in TO-252, and its spec rows include a low-voltage R_DS(on) line the NCE original doesn't publish. A clone that is better on paper at 5V is a different die, not a better one. Match the marking and check that the only on-resistance row is the 10V one.

Q10: What does the v1.1 datasheet revision change?

A: The sheet we work from is v1.1, and the numbers here come from it. Cross-reference databases and distributor tables often carry older or rounded figures — dissipation ratings that are double the sheet's, or typical resistances presented as guaranteed maxima. When a listing and the PDF disagree, the PDF with the revision number on it wins.